ENEGEP2023_TN_WG_404_1987_45438
EMOTIONS DETECTION IN SOCIAL MEDIA POSTS
ÁREA: 06. ENGENHARIA ORGANIZACIONAL
SUBÁREA: 06.4. Gestão da Informação
RESUMO:
With the proliferation of the internet, the number of users expressing their opinions and perceptions about different entities on social media has grown. So far, research on identifying emotions in text using natural language processing has mainly focused on the English language, using Machine Learning approaches. This results in a limited understanding of the phenomenon, especially in other languages such as Brazilian Portuguese. This study aims to detect emotions in Twitter posts related to a Brazilian soccer team using a novel natural language processing approach. To this end, a test corpus composed by tweets were studied in detail to create simple rules for how a human would perform this task and to create a taxonomy of emotions specific to soccer fans. This logic was implemented in NLP++, a computer language specifically designed for encoding the way humans process text that differs from common machine learning techniques and requires only a small number of examples to be implemented. As an output, a graph was generated showing the percentages of each emotion at each moment of the soccer match using a data visualization library from HPCC Systems. The results of this study seem to indicate a viable approach to reproduce human thinking in the detection of emotions on a large scale, and NLP++ proved to be a powerful tool for dealing with this challenge in the natural language processing field.
PALAVRAS-CHAVE:
big data, natural language processing, emotion analysis, soccer team, social media.
DOI: 10.14488/enegep2023_tn_wg_404_1987_45438
SUBÁREA: 06.4. Gestão da Informação
RESUMO:
With the proliferation of the internet, the number of users expressing their opinions and perceptions about different entities on social media has grown. So far, research on identifying emotions in text using natural language processing has mainly focused on the English language, using Machine Learning approaches. This results in a limited understanding of the phenomenon, especially in other languages such as Brazilian Portuguese. This study aims to detect emotions in Twitter posts related to a Brazilian soccer team using a novel natural language processing approach. To this end, a test corpus composed by tweets were studied in detail to create simple rules for how a human would perform this task and to create a taxonomy of emotions specific to soccer fans. This logic was implemented in NLP++, a computer language specifically designed for encoding the way humans process text that differs from common machine learning techniques and requires only a small number of examples to be implemented. As an output, a graph was generated showing the percentages of each emotion at each moment of the soccer match using a data visualization library from HPCC Systems. The results of this study seem to indicate a viable approach to reproduce human thinking in the detection of emotions on a large scale, and NLP++ proved to be a powerful tool for dealing with this challenge in the natural language processing field.
PALAVRAS-CHAVE:
big data, natural language processing, emotion analysis, soccer team, social media.
DOI: 10.14488/enegep2023_tn_wg_404_1987_45438
AUTORES:
PEDRO LIMA RODRIGUES
UNIVERSIDADE DE SÃO PAULO
RENATO DE OLIVEIRA MORAES
DEPARTAMENTO DE ENGENHARIA DE PRODUÇÃO DA ESCOLA POLITÉCNICA
HUGO MARTINELLI WATANUKI
DEPARTAMENTO DE ENGENHARIA DE PRODUÇÃO DA ESCOLA POLITÉCNICA
DAVID DE HILSTER
LEXISNEXIS
PEDRO LIMA RODRIGUES
UNIVERSIDADE DE SÃO PAULO
RENATO DE OLIVEIRA MORAES
DEPARTAMENTO DE ENGENHARIA DE PRODUÇÃO DA ESCOLA POLITÉCNICA
HUGO MARTINELLI WATANUKI
DEPARTAMENTO DE ENGENHARIA DE PRODUÇÃO DA ESCOLA POLITÉCNICA
DAVID DE HILSTER
LEXISNEXIS
|